[HACKERS] The cost of visibillity testing? (gin-search)

2010-12-21 Thread Jesper Krogh
Hi Hackers. I have a feeling that GIN is cheating on the visibillity checks: test=# set enable_seqscan = off; SET Time: 0.129 ms test=# select count(id) from fts_test where fts @@ to_tsquery('core'); count 158827 (1 row) Time: 95.530 ms test=# explain select count(id) from fts_test

Re: [HACKERS] The cost of visibillity testing? (gin-search)

2010-12-21 Thread Heikki Linnakangas
On 21.12.2010 21:25, Jesper Krogh wrote: The first query should have the cost of the GIN-search + visibillity-test of 158K tuples, the latter should have the cost of visibillity-testing 168K tuples. If we set the cost of actually searching GIN to 0 then the gin-search - visibillity costs:

Re: [HACKERS] The cost of visibillity testing? (gin-search)

2010-12-21 Thread Andres Freund
On Tuesday 21 December 2010 20:25:16 Jesper Krogh wrote: What have I missed in the logic? A reproducible testcase ;-) Andres -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers

Re: [HACKERS] The cost of visibillity testing? (gin-search)

2010-12-21 Thread Tom Lane
Heikki Linnakangas heikki.linnakan...@enterprisedb.com writes: On 21.12.2010 21:25, Jesper Krogh wrote: Or is a Bitmap Heap Scan simply 3 times faster than a Seq-scan for visibillity-testing? It certainly shouldn't be. What have I missed in the logic? Perhaps you have a lot of empty space

Re: [HACKERS] The cost of visibillity testing? (gin-search)

2010-12-21 Thread Jesper Krogh
On 2010-12-21 21:28, Andres Freund wrote: On Tuesday 21 December 2010 20:25:16 Jesper Krogh wrote: What have I missed in the logic? A reproducible testcase ;-) Yes, I did a complete dump/restore of the dataset and the numbers looked like expected. So table bloat seems to be the